[Paper Review] A lower bound on the overhead of quantum error correction in low dimensions

Nouédyn Baspin, Omar Fawzi+1 more|arXiv (Cornell University)|Feb 8, 2023

Quantum Computing Algorithms and Architecture4 citations

TL;DR

This paper establishes a fundamental lower bound on the space overhead required for quantum error correction in low-dimensional quantum architectures using only geometrically local operations. It proves that in 2D, the overhead scales as Ω(√log(1/δ)) for logical error rate δ, even with arbitrary classical computation, due to limitations in entanglement generation and locality constraints.

ABSTRACT

We show that a quantum architecture with an error correction procedure limited to geometrically local operations incurs an overhead that grows with the system size, even if arbitrary error-free classical computation is allowed. In particular, we prove that in order to operate a quantum error correcting code in 2D at a logical error rate of $δ$, a space overhead of $Ω(\sqrt{\log(1/δ)})$ is needed for any constant depolarizing noise $p > 0$.

Motivation & Objective

• To determine the minimal space overhead required for quantum error correction in low-dimensional systems with only geometrically local operations.
• To investigate whether arbitrary classical computation can overcome the limitations imposed by geometric locality in quantum error correction.
• To establish a lower bound on the overhead that scales with the inverse logical error rate and system dimensionality.
• To formalize the connection between error correction performance and the ability to generate entanglement in spatially constrained architectures.
• To generalize the bound beyond 2D to D-dimensional systems, showing a dependence on the dimension D through a geometric function g_geom.

Proposed method

• Uses a framework of separable operations between quantum and classical systems, allowing arbitrary classical computation but restricting quantum operations to local interactions.
• Applies a partitioning argument on the physical qubit register A into subsets Γ_i with bounded boundary size |∂Γ_i|, leveraging isoperimetric inequalities in D dimensions.
• Employs entanglement measures E_R(Λ_i : ḞΛ_i) in the encoded state to quantify information flow and error resilience.
• Applies a quantum error correction fidelity condition (Eq. 2) to relate logical error rate δ to the performance of recovery channels.
• Derives a lower bound on the number of physical qubits m via a chain of inequalities involving the partition size, boundary size, and error parameters.
• Uses geometric constants c1(D), c2(D) and the parameter f = log_p(δ) to express the overhead in terms of D and δ, leading to the final bound.

Experimental results

Research questions

• RQ1 What is the minimal space overhead required to achieve a target logical error rate δ in a 2D quantum architecture with only local operations?
• RQ2 Can arbitrary classical computation compensate for the limitations of geometric locality in quantum error correction?
• RQ3 How does the overhead scale with the logical error rate δ and the noise parameter p in D-dimensional systems?
• RQ4 What is the fundamental trade-off between locality, entanglement generation, and error correction performance?
• RQ5 Does the geometric structure of the system impose an intrinsic lower bound on the overhead, independent of code design?

Key findings

• In 2D, the space overhead for quantum error correction with local operations and arbitrary classical computation scales as Ω(√log(1/δ)) for any constant depolarizing noise p > 0.
• The bound generalizes to D dimensions, yielding an overhead of Ω((log(1/δ))^{1/D}), showing a diminishing return with increasing dimensionality.
• The overhead is lower bounded by Ω(m^{c/2}) when the logical error rate δ decays exponentially with system size m.
• The result holds even when classical computation is free, proving that locality in quantum operations is a fundamental bottleneck.
• The bound is derived using entanglement measures and geometric partitioning, highlighting the role of non-locality in error correction performance.
• The analysis applies to any architecture with limited entanglement generation speed, linking error correction capability directly to entanglement dynamics.